Query logs and traces with SQL in Google Cloud Observability Analytics

Troubleshooting distributed systems requires stitching together logs and traces, but these data sources have historically lived in separate tools with separate query interfaces. This creates friction when engineers need to answer systemic questions like “which microservice caused a 5-second checkout slowdown” or “which tool call fails most frequently across thousands of AI agent runs.” Google Cloud announces Observability Analytics (formerly Log Analytics) and SQL support in Cloud Trace to unify these telemetry streams into a single BigQuery-backed analytical surface.

The core technical move is bringing BigQuery and SQL directly into Cloud Observability, so you can JOIN log entries and trace spans in place without exporting or duplicating data. This enables aggregate queries across millions of spans and log events — for example, ranking AI agent tools by failure rate and P95 latency over the last 7 days, or identifying the top 10 customers experiencing the worst latency by joining trace attributes with application log identifiers. A new Observability API also makes telemetry available as linked BigQuery datasets, allowing AI agents or analytical workloads to query the data programmatically via standard BigQuery APIs.

For serious builders, the takeaway is that operational data silos are no longer a necessity. If you already run on Google Cloud, you can now treat logs and traces as a single BigQuery dataset and run analytical SQL across both, which is especially useful for AI agent observability at scale — identifying flaky tool calls, slow endpoints, or latency bottlenecks tied to specific customer segments. The Log and Trace explorers remain the right tools for drilling into individual incidents, but Observability Analytics is where you go for the aggregate view: systemic patterns, business correlation, and programmatic access via AI agents.